[Intel-gfx] [PATCH] drm/i915: Introduce for_each_ring() macro

Daniel Vetter daniel at ffwll.ch
Fri May 11 11:00:12 CEST 2012


On Thu, May 10, 2012 at 03:39:32PM -0700, Ben Widawsky wrote:
> On Thu, 10 May 2012 22:21:50 +0100
> Chris Wilson <chris at chris-wilson.co.uk> wrote:
> 
> > In many places we wish to iterate over the rings associated with the
> > GPU, so refactor them to use a common macro.
> > 
> > Along the way, there are a few code removals that should be side-effect
> > free and some rearrangement which should only have a cosmetic impact,
> > such as error-state.
> > 
> > Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Reviewed-by: Ben Widawsky <ben at bwidawsk.net> (if last hunk is removed
> and commit messages is amended)
> > ---
> >  drivers/gpu/drm/i915/i915_debugfs.c   |    9 ++---
> >  drivers/gpu/drm/i915/i915_drv.c       |   10 ++---
> >  drivers/gpu/drm/i915/i915_drv.h       |    9 +++--
> >  drivers/gpu/drm/i915/i915_gem.c       |   33 +++++++---------
> >  drivers/gpu/drm/i915/i915_gem_evict.c |   14 +++----
> >  drivers/gpu/drm/i915/i915_irq.c       |   69 +++++++++++++--------------------
> >  drivers/gpu/drm/i915/intel_pm.c       |    6 ++-
> >  7 files changed, 66 insertions(+), 84 deletions(-)
> > 
> > diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
> > index 950f72a..eb2b3c2 100644
> > --- a/drivers/gpu/drm/i915/i915_debugfs.c
> > +++ b/drivers/gpu/drm/i915/i915_debugfs.c
> > @@ -699,6 +699,7 @@ static int i915_error_state(struct seq_file *m, void *unused)
> >  	struct drm_device *dev = error_priv->dev;
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> >  	struct drm_i915_error_state *error = error_priv->error;
> > +	struct intel_ring_buffer *ring;
> >  	int i, j, page, offset, elt;
> >  
> >  	if (!error) {
> > @@ -706,7 +707,6 @@ static int i915_error_state(struct seq_file *m, void *unused)
> >  		return 0;
> >  	}
> >  
> > -
> >  	seq_printf(m, "Time: %ld s %ld us\n", error->time.tv_sec,
> >  		   error->time.tv_usec);
> >  	seq_printf(m, "PCI ID: 0x%04x\n", dev->pci_device);
> > @@ -722,11 +722,8 @@ static int i915_error_state(struct seq_file *m, void *unused)
> >  		seq_printf(m, "DONE_REG: 0x%08x\n", error->done_reg);
> >  	}
> >  
> > -	i915_ring_error_state(m, dev, error, RCS);
> > -	if (HAS_BLT(dev))
> > -		i915_ring_error_state(m, dev, error, BCS);
> > -	if (HAS_BSD(dev))
> > -		i915_ring_error_state(m, dev, error, VCS);
> > +	for_each_ring(ring, dev_priv, i)
> > +		i915_ring_error_state(m, dev, error, i);
> >  
> >  	if (error->active_bo)
> >  		print_error_buffers(m, "Active",
> > diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> > index 5898be9..2b7967d 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.c
> > +++ b/drivers/gpu/drm/i915/i915_drv.c
> > @@ -893,15 +893,15 @@ int i915_reset(struct drm_device *dev)
> >  	 */
> >  	if (drm_core_check_feature(dev, DRIVER_MODESET) ||
> >  			!dev_priv->mm.suspended) {
> > +		struct intel_ring_buffer *ring;
> > +		int i;
> > +
> >  		dev_priv->mm.suspended = 0;
> >  
> >  		i915_gem_init_swizzling(dev);
> >  
> > -		dev_priv->ring[RCS].init(&dev_priv->ring[RCS]);
> > -		if (HAS_BSD(dev))
> > -		    dev_priv->ring[VCS].init(&dev_priv->ring[VCS]);
> > -		if (HAS_BLT(dev))
> > -		    dev_priv->ring[BCS].init(&dev_priv->ring[BCS]);
> > +		for_each_ring(ring, dev_priv, i)
> > +			ring->init(ring, ring->obj);
> >  
> >  		i915_gem_init_ppgtt(dev);
> >  
> > diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h
> > index 83a557c..c2e0732 100644
> > --- a/drivers/gpu/drm/i915/i915_drv.h
> > +++ b/drivers/gpu/drm/i915/i915_drv.h
> > @@ -410,9 +410,7 @@ typedef struct drm_i915_private {
> >  #define DRM_I915_HANGCHECK_PERIOD 1500 /* in ms */
> >  	struct timer_list hangcheck_timer;
> >  	int hangcheck_count;
> > -	uint32_t last_acthd;
> > -	uint32_t last_acthd_bsd;
> > -	uint32_t last_acthd_blt;
> > +	uint32_t last_acthd[I915_NUM_RINGS];
> >  	uint32_t last_instdone;
> >  	uint32_t last_instdone1;
> >  
> > @@ -820,6 +818,11 @@ typedef struct drm_i915_private {
> >  	struct drm_property *force_audio_property;
> >  } drm_i915_private_t;
> >  
> > +/* Iterate over initialised rings */
> > +#define for_each_ring(ring__, dev_priv__, i__) \
> > +	for ((i__) = 0; (i__) < I915_NUM_RINGS; (i__)++) \
> > +		if (((ring__) = &(dev_priv__)->ring[(i__)])->obj)
> > +
> >  enum hdmi_force_audio {
> >  	HDMI_AUDIO_OFF_DVI = -2,	/* no aux data for HDMI-DVI converter */
> >  	HDMI_AUDIO_OFF,			/* force turn off HDMI audio */
> 
> I'm somewhat disappointed with using ring->obj here. But I do not have a
> better suggestion.

I think we could wrap the ring->obj checking into intel_ring_initialized
and adjust the if checke like

	if (((ring_) = &(dev_priv__)->ring[(i__)]) && intel_ring_initialized (ring__))

The follow-on patch to use the ring->obj check more would then also be
self-documenting.

> > diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> > index 44a5f24..99961ce 100644
> > --- a/drivers/gpu/drm/i915/i915_gem.c
> > +++ b/drivers/gpu/drm/i915/i915_gem.c
> > @@ -1655,10 +1655,11 @@ void i915_gem_reset(struct drm_device *dev)
> >  {
> >  	struct drm_i915_private *dev_priv = dev->dev_private;
> >  	struct drm_i915_gem_object *obj;
> > +	struct intel_ring_buffer *ring;
> >  	int i;
> >  
> > -	for (i = 0; i < I915_NUM_RINGS; i++)
> > -		i915_gem_reset_ring_lists(dev_priv, &dev_priv->ring[i]);
> > +	for_each_ring(ring, dev_priv, i)
> > +		i915_gem_reset_ring_lists(dev_priv, ring);
> >  
> >  	/* Remove anything from the flushing lists. The GPU cache is likely
> >  	 * to be lost on reset along with the data, so simply move the
> > @@ -1763,10 +1764,11 @@ void
> >  i915_gem_retire_requests(struct drm_device *dev)
> >  {
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> > +	struct intel_ring_buffer *ring;
> >  	int i;
> >  
> > -	for (i = 0; i < I915_NUM_RINGS; i++)
> > -		i915_gem_retire_requests_ring(&dev_priv->ring[i]);
> > +	for_each_ring(ring, dev_priv, i)
> > +		i915_gem_retire_requests_ring(ring);
> >  }
> >  
> >  static void
> > @@ -1774,6 +1776,7 @@ i915_gem_retire_work_handler(struct work_struct *work)
> >  {
> >  	drm_i915_private_t *dev_priv;
> >  	struct drm_device *dev;
> > +	struct intel_ring_buffer *ring;
> >  	bool idle;
> >  	int i;
> >  
> > @@ -1793,9 +1796,7 @@ i915_gem_retire_work_handler(struct work_struct *work)
> >  	 * objects indefinitely.
> >  	 */
> >  	idle = true;
> > -	for (i = 0; i < I915_NUM_RINGS; i++) {
> > -		struct intel_ring_buffer *ring = &dev_priv->ring[i];
> > -
> > +	for_each_ring(ring, dev_priv, i) {
> >  		if (!list_empty(&ring->gpu_write_list)) {
> >  			struct drm_i915_gem_request *request;
> >  			int ret;
> > @@ -2137,10 +2138,11 @@ static int i915_ring_idle(struct intel_ring_buffer *ring)
> >  int i915_gpu_idle(struct drm_device *dev)
> >  {
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> > +	struct intel_ring_buffer *ring;
> >  	int ret, i;
> >  
> >  	/* Flush everything onto the inactive list. */
> > -	for (i = 0; i < I915_NUM_RINGS; i++) {
> > +	for_each_ring(ring, dev_priv, i) {
> >  		ret = i915_ring_idle(&dev_priv->ring[i]);
> >  		if (ret)
> >  			return ret;
> > @@ -3463,9 +3465,7 @@ void i915_gem_init_ppgtt(struct drm_device *dev)
> >  		/* GFX_MODE is per-ring on gen7+ */
> >  	}
> >  
> > -	for (i = 0; i < I915_NUM_RINGS; i++) {
> > -		ring = &dev_priv->ring[i];
> > -
> > +	for_each_ring(ring, dev_priv, i) {
> >  		if (INTEL_INFO(dev)->gen >= 7)
> >  			I915_WRITE(RING_MODE_GEN7(ring),
> >  				   _MASKED_BIT_ENABLE(GFX_PPGTT_ENABLE));
> > @@ -3581,10 +3581,11 @@ void
> >  i915_gem_cleanup_ringbuffer(struct drm_device *dev)
> >  {
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> > +	struct intel_ring_buffer *ring;
> >  	int i;
> >  
> > -	for (i = 0; i < I915_NUM_RINGS; i++)
> > -		intel_cleanup_ring_buffer(&dev_priv->ring[i]);
> > +	for_each_ring(ring, dev_priv, i)
> > +		intel_cleanup_ring_buffer(ring);
> >  }
> >  
> >  int
> > @@ -3592,7 +3593,7 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
> >  		       struct drm_file *file_priv)
> >  {
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> > -	int ret, i;
> > +	int ret;
> >  
> >  	if (drm_core_check_feature(dev, DRIVER_MODESET))
> >  		return 0;
> > @@ -3614,10 +3615,6 @@ i915_gem_entervt_ioctl(struct drm_device *dev, void *data,
> >  	BUG_ON(!list_empty(&dev_priv->mm.active_list));
> >  	BUG_ON(!list_empty(&dev_priv->mm.flushing_list));
> >  	BUG_ON(!list_empty(&dev_priv->mm.inactive_list));
> > -	for (i = 0; i < I915_NUM_RINGS; i++) {
> > -		BUG_ON(!list_empty(&dev_priv->ring[i].active_list));
> > -		BUG_ON(!list_empty(&dev_priv->ring[i].request_list));
> > -	}
> >  	mutex_unlock(&dev->struct_mutex);
> >  
> >  	ret = drm_irq_install(dev);
> 
> I'm not opposed to this change, but it would have probably been better
> to not remove the BUG_ONs.

Shouldn't the ring->obj check dtrt here and we could just convert that
BUG_ON loop to for_each_ring?

> > index 3bcf045..ae7c24e 100644
> > --- a/drivers/gpu/drm/i915/i915_gem_evict.c
> > +++ b/drivers/gpu/drm/i915/i915_gem_evict.c
> > @@ -168,7 +168,7 @@ i915_gem_evict_everything(struct drm_device *dev, bool purgeable_only)
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> >  	struct drm_i915_gem_object *obj, *next;
> >  	bool lists_empty;
> > -	int ret,i;
> > +	int ret;
> >  
> >  	lists_empty = (list_empty(&dev_priv->mm.inactive_list) &&
> >  		       list_empty(&dev_priv->mm.flushing_list) &&
> > @@ -178,17 +178,13 @@ i915_gem_evict_everything(struct drm_device *dev, bool purgeable_only)
> >  
> >  	trace_i915_gem_evict_everything(dev, purgeable_only);
> >  
> > -	ret = i915_gpu_idle(dev);
> > -	if (ret)
> > -		return ret;
> > -
> >  	/* The gpu_idle will flush everything in the write domain to the
> >  	 * active list. Then we must move everything off the active list
> >  	 * with retire requests.
> >  	 */
> > -	for (i = 0; i < I915_NUM_RINGS; i++)
> > -		if (WARN_ON(!list_empty(&dev_priv->ring[i].gpu_write_list)))
> > -			return -EBUSY;
> > +	ret = i915_gpu_idle(dev);
> > +	if (ret)
> > +		return ret;
> >  
> >  	i915_gem_retire_requests(dev);
> >  
> 
> Sneaky way to get rid of the warning you never liked... I won't forget
> this.

I concur that killing that warnings is sneaky ;-) Can't we just keep them
- the ring->obj check should do the right thing? Or what exactly do I
miss?

> > @@ -203,5 +199,5 @@ i915_gem_evict_everything(struct drm_device *dev, bool purgeable_only)
> >  		}
> >  	}
> >  
> > -	return ret;
> > +	return 0;
> >  }
> > diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
> > index 2a062d7..cc4a633 100644
> > --- a/drivers/gpu/drm/i915/i915_irq.c
> > +++ b/drivers/gpu/drm/i915/i915_irq.c
> > @@ -1022,15 +1022,11 @@ static void i915_gem_record_rings(struct drm_device *dev,
> >  				  struct drm_i915_error_state *error)
> >  {
> >  	struct drm_i915_private *dev_priv = dev->dev_private;
> > +	struct intel_ring_buffer *ring;
> >  	struct drm_i915_gem_request *request;
> >  	int i, count;
> >  
> > -	for (i = 0; i < I915_NUM_RINGS; i++) {
> > -		struct intel_ring_buffer *ring = &dev_priv->ring[i];
> > -
> > -		if (ring->obj == NULL)
> > -			continue;
> > -
> > +	for_each_ring(ring, dev_priv, i) {
> >  		i915_record_ring_state(dev, error, ring);
> >  
> >  		error->ring[i].batchbuffer =
> > @@ -1295,6 +1291,8 @@ static void i915_report_and_clear_eir(struct drm_device *dev)
> >  void i915_handle_error(struct drm_device *dev, bool wedged)
> >  {
> >  	struct drm_i915_private *dev_priv = dev->dev_private;
> > +	struct intel_ring_buffer *ring;
> > +	int i;
> >  
> >  	i915_capture_error_state(dev);
> >  	i915_report_and_clear_eir(dev);
> > @@ -1306,11 +1304,8 @@ void i915_handle_error(struct drm_device *dev, bool wedged)
> >  		/*
> >  		 * Wakeup waiting processes so they don't hang
> >  		 */
> > -		wake_up_all(&dev_priv->ring[RCS].irq_queue);
> > -		if (HAS_BSD(dev))
> > -			wake_up_all(&dev_priv->ring[VCS].irq_queue);
> > -		if (HAS_BLT(dev))
> > -			wake_up_all(&dev_priv->ring[BCS].irq_queue);
> > +		for_each_ring(ring, dev_priv, i)
> > +			wake_up_all(&ring->irq_queue);
> >  	}
> >  
> >  	queue_work(dev_priv->wq, &dev_priv->error_work);
> > @@ -1515,11 +1510,6 @@ ring_last_seqno(struct intel_ring_buffer *ring)
> >  
> >  static bool i915_hangcheck_ring_idle(struct intel_ring_buffer *ring, bool *err)
> >  {
> > -	/* We don't check whether the ring even exists before calling this
> > -	 * function. Hence check whether it's initialized. */
> > -	if (ring->obj == NULL)
> > -		return true;
> > -
> >  	if (list_empty(&ring->request_list) ||
> >  	    i915_seqno_passed(ring->get_seqno(ring), ring_last_seqno(ring))) {
> >  		/* Issue a wake-up to catch stuck h/w. */
> > @@ -1553,26 +1543,25 @@ static bool i915_hangcheck_hung(struct drm_device *dev)
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> >  
> >  	if (dev_priv->hangcheck_count++ > 1) {
> > +		bool hung = true;
> > +
> >  		DRM_ERROR("Hangcheck timer elapsed... GPU hung\n");
> >  		i915_handle_error(dev, true);
> >  
> >  		if (!IS_GEN2(dev)) {
> > +			struct intel_ring_buffer *ring;
> > +			int i;
> > +
> >  			/* Is the chip hanging on a WAIT_FOR_EVENT?
> >  			 * If so we can simply poke the RB_WAIT bit
> >  			 * and break the hang. This should work on
> >  			 * all but the second generation chipsets.
> >  			 */
> > -			if (kick_ring(&dev_priv->ring[RCS]))
> > -				return false;
> > -
> > -			if (HAS_BSD(dev) && kick_ring(&dev_priv->ring[VCS]))
> > -				return false;
> > -
> > -			if (HAS_BLT(dev) && kick_ring(&dev_priv->ring[BCS]))
> > -				return false;
> > +			for_each_ring(ring, dev_priv, i)
> > +				hung &= !kick_ring(ring);
> >  		}
> >  
> > -		return true;
> > +		return hung;
> >  	}
> >  
> >  	return false;
> 
> It's worth noting in the commit message that there is a functional
> change here. Before we'd exit out on the first ring kicked, now we
> enumerate over all rings regardless.

I agree that's noteworthy (and poking all rings makes more sense anyway).

> I'm still a fan of ripping out the kicks completely.

We can't, it helps to get us out of broken userspace on gen3 (where we
lack gpu reset support).

> > @@ -1588,16 +1577,23 @@ void i915_hangcheck_elapsed(unsigned long data)
> >  {
> >  	struct drm_device *dev = (struct drm_device *)data;
> >  	drm_i915_private_t *dev_priv = dev->dev_private;
> > -	uint32_t acthd, instdone, instdone1, acthd_bsd, acthd_blt;
> > -	bool err = false;
> > +	uint32_t acthd[I915_NUM_RINGS], instdone, instdone1;
> > +	struct intel_ring_buffer *ring;
> > +	bool err = false, idle;
> > +	int i;
> >  
> >  	if (!i915_enable_hangcheck)
> >  		return;
> >  
> > +	memset(acthd, 0, sizeof(acthd));
> > +	idle = true;
> > +	for_each_ring(ring, dev_priv, i) {
> > +	    idle &= i915_hangcheck_ring_idle(ring, &err);
> > +	    acthd[i] = intel_ring_get_active_head(ring);
> > +	}
> > +
> >  	/* If all work is done then ACTHD clearly hasn't advanced. */
> > -	if (i915_hangcheck_ring_idle(&dev_priv->ring[RCS], &err) &&
> > -	    i915_hangcheck_ring_idle(&dev_priv->ring[VCS], &err) &&
> > -	    i915_hangcheck_ring_idle(&dev_priv->ring[BCS], &err)) {
> > +	if (idle) {
> >  		if (err) {
> >  			if (i915_hangcheck_hung(dev))
> >  				return;
> 
> Similar functional change as previous: we do all rings no matter what.
> 
> > @@ -1616,15 +1612,8 @@ void i915_hangcheck_elapsed(unsigned long data)
> >  		instdone = I915_READ(INSTDONE_I965);
> >  		instdone1 = I915_READ(INSTDONE1);
> >  	}
> > -	acthd = intel_ring_get_active_head(&dev_priv->ring[RCS]);
> > -	acthd_bsd = HAS_BSD(dev) ?
> > -		intel_ring_get_active_head(&dev_priv->ring[VCS]) : 0;
> > -	acthd_blt = HAS_BLT(dev) ?
> > -		intel_ring_get_active_head(&dev_priv->ring[BCS]) : 0;
> >  
> > -	if (dev_priv->last_acthd == acthd &&
> > -	    dev_priv->last_acthd_bsd == acthd_bsd &&
> > -	    dev_priv->last_acthd_blt == acthd_blt &&
> > +	if (memcmp(dev_priv->last_acthd, acthd, sizeof(acthd)) == 0 &&
> >  	    dev_priv->last_instdone == instdone &&
> >  	    dev_priv->last_instdone1 == instdone1) {
> >  		if (i915_hangcheck_hung(dev))
> > @@ -1632,9 +1621,7 @@ void i915_hangcheck_elapsed(unsigned long data)
> >  	} else {
> >  		dev_priv->hangcheck_count = 0;
> >  
> > -		dev_priv->last_acthd = acthd;
> > -		dev_priv->last_acthd_bsd = acthd_bsd;
> > -		dev_priv->last_acthd_blt = acthd_blt;
> > +		memcpy(dev_priv->last_acthd, acthd, sizeof(acthd));
> >  		dev_priv->last_instdone = instdone;
> >  		dev_priv->last_instdone1 = instdone1;
> >  	}
> > diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
> > index 8f8d1da..438ff29 100644
> > --- a/drivers/gpu/drm/i915/intel_pm.c
> > +++ b/drivers/gpu/drm/i915/intel_pm.c
> > @@ -2326,6 +2326,7 @@ int intel_enable_rc6(const struct drm_device *dev)
> >  
> >  void gen6_enable_rps(struct drm_i915_private *dev_priv)
> >  {
> > +	struct intel_ring_buffer *ring;
> >  	u32 rp_state_cap = I915_READ(GEN6_RP_STATE_CAP);
> >  	u32 gt_perf_status = I915_READ(GEN6_GT_PERF_STATUS);
> >  	u32 pcu_mbox, rc6_mask = 0;
> > @@ -2360,8 +2361,8 @@ void gen6_enable_rps(struct drm_i915_private *dev_priv)
> >  	I915_WRITE(GEN6_RC_EVALUATION_INTERVAL, 125000);
> >  	I915_WRITE(GEN6_RC_IDLE_HYSTERSIS, 25);
> >  
> > -	for (i = 0; i < I915_NUM_RINGS; i++)
> > -		I915_WRITE(RING_MAX_IDLE(dev_priv->ring[i].mmio_base), 10);
> > +	for_each_ring(ring, dev_priv, i)
> > +		I915_WRITE(RING_MAX_IDLE(ring->mmio_base), 10);
> >  
> >  	I915_WRITE(GEN6_RC_SLEEP, 0);
> >  	I915_WRITE(GEN6_RC1e_THRESHOLD, 1000);
> > @@ -3048,6 +3049,7 @@ EXPORT_SYMBOL_GPL(i915_gpu_lower);
> >  bool i915_gpu_busy(void)
> >  {
> >  	struct drm_i915_private *dev_priv;
> > +	struct intel_ring_buffer *ring;
> >  	bool ret = false;
> >  
> >  	spin_lock(&mchdev_lock);
> 
> 
> There is still a hunk fail here.
> 
> -- 
> Ben Widawsky, Intel Open Source Technology Center
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Mail: daniel at ffwll.ch
Mobile: +41 (0)79 365 57 48



More information about the Intel-gfx mailing list